Planet Air

Analysis exploring air quality inequality across U.S. counties and territories using EPA AQI data.

Project by: Kelly Chang, Jennie Zheng, Emily Chi

How is exposure to unhealthy air quality distributed across U.S. counties and states, and what regional patterns emerge in air pollution risk over time?

Why We Care

Identifying where unhealthy air exposure is concentrated is important for understanding potential public health risks, as repeated exposure to poor air quality is associated with respiratory and cardiovascular problems. Uneven exposure patterns across regions raise concerns about fairness in access to clean air and highlight areas where further investigation may be needed. These findings also support more informed decision making by showing how exposure differs across geographic scales and over time.

States that Experience the Most Unhealthy or Hazardous Days

County-Level Patterns in Unhealthy Air Exposure

The county-level map shows that unhealthy or hazardous air quality days are unevenly distributed across the United States. While many counties experience few unhealthy days, there are clear clusters of higher exposure, particularly in parts of the Southwest and Western U.S., indicating that air quality risk is highly localized. In this analysis, unhealthy air quality days were defined as the sum of Unhealthy for Sensitive Groups, Unhealthy, Very Unhealthy, and Hazardous days. To make these patterns easier to interpret, counties were grouped into bins based on the number of unhealthy or hazardous days (for example, 1–5 days, 6–15 days, and 30+ days). Using bins instead of raw day counts reduces visual noise and prevents small numerical differences from appearing more meaningful than they are, allowing regional patterns and high-exposure clusters to be identified more clearly than if exact values were shown.

State-Level Exposure: Overall Impact vs. Average County Risk

The two state-level bar charts show that which states experience the most unhealthy air days depends on how exposure is measured. The total unhealthy days chart highlights states such as California and Texas, reflecting the overall cumulative burden across counties, but this measure is influenced by state size and the number of counties. The average unhealthy days per county chart provides a complementary view by highlighting states where counties experience consistently higher exposure on average, such as North Dakota and Arizona. Both graphs are included because they capture different dimensions of exposure: overall exposure and typical county-level exposure.

Limitations

These visualizations show where unhealthy air exposure is concentrated but do not explain which specific pollutants drive that exposure or how air quality has changed over time. Those questions are addressed in other parts of the analysis. Data availability also varies across counties, with many central U.S. counties showing limited or missing AQI data, likely due to fewer monitoring stations. This may cause air quality risks in rural areas to be underestimated. In addition, U.S. territories such as Puerto Rico and the U.S. Virgin Islands are not included due to data and geographic matching constraints.

How does exposure to specific pollutants (PM2.5, Ozone, NO₂, PM10, CO) vary across counties?

Key Findings

This map visualizes the dominant air pollutant in each U.S. county based on the pollutant exposure days recorded in the EPA AQI dataset. Each county is color-coded by the pollutant with the greatest presencem, such as PM2.5, Ozone, or NO₂ allowing viewers to quickly identify regional pollution patterns.

The spatial distribution of dominant pollutants reveals clear regional differences in air quality exposure across the United States. Ozone appears heavily concentrated in the Northeast and stretches across much of the Midwest and Southeast.

In contrast, PM2.5 is more widely dispersed across both western and eastern states, appearing frequently along the West Coast as well as across parts of the South and Midwest.

Other pollutants like PM10, NO₂, and CO appear less dominant overall and occur more sporadically, suggesting localized rather than regional trends.

Limitations

While the map highlights regional patterns in dominant pollutants, it simplifies complex air quality data into a single “top pollutant” per county. This means counties with similar colors may still have very different overall pollution severity or health risk levels that are not visible in the visualization. For example, a county labeled as ozone-dominant may still experience significant particulate matter exposure that is hidden by the categorical coloring. Additionally, the visualization represents dominance by number of exposure days, not pollutant concentration or toxicity. Some pollutants may appear less frequently but have stronger health impacts, which this map does not capture. As a result, the graph is best interpreted as a comparison of relative exposure patterns rather than a direct measure of environmental harm.

How AOI Trend Change Across Different States/Territories in The Last 10 Years

Changes in Air Quality Across U.S. States and Territories (2015–2025)

This visualization shows how air quality changed across U.S. states and territories between 2015 and 2025, calculated as the difference between values in 2025 and 2015 (2025 − 2015). Negative values indicate improvement, while positive values indicate worsening air quality. These changes were computed by directly comparing reported AQI metrics for the two years.

Improvements (2015–2025)

The most substantial improvement occurred in the U.S. Virgin Islands, where the metric decreased from 43.00 in 2015 to 18.00 in 2025, resulting in a change of −25.00. Hawaii also showed a notable improvement, declining from 38.00 to 27.66 (−10.34), followed by Montana, which decreased from 32.94 to 23.17 (−9.77). These large decreases indicate meaningful long-term gains in air quality, particularly in island regions and parts of the western United States.

However, the Virgin Islands result should be interpreted with caution. U.S. territories are not displayed on the state-level map and typically have fewer air quality monitoring locations than U.S. states. As a result, the magnitude of the observed change may be influenced by limited data coverage or changes in monitoring practices over time, even though the numerical improvement is large.

Worsening Conditions (2015–2025)

In contrast, the greatest worsening was observed in Oklahoma, where values increased from 38.61 in 2015 to 43.75 in 2025, an increase of +5.14. South Carolina experienced a similar rise from 39.26 to 43.77 (+4.51), followed by North Dakota, which increased from 35.50 to 39.78 (+4.28). These increases reflect declining air quality over the same period in several Southern and Midwestern states.

Overall Pattern

Taken together, the data reveal a clear regional pattern: western states and U.S. territories tend to show the largest improvements in air quality between 2015 and 2025, while many central and southeastern states exhibit worsening conditions. This pattern highlights that progress in air quality has been uneven across the United States over the past decade.

Limitations

This time-trend analysis measures changes in air quality by comparing values from 2015 and 2025, calculated as the difference between the two years. While this approach captures long-term change, it does not reflect year-to-year variability or short-term pollution events that may have occurred within the decade.

Results may also be influenced by uneven monitoring coverage across states and territories. Some regions, particularly U.S. territories and more rural areas, have fewer air quality monitoring stations, which can affect the stability and representativeness of observed changes. For example, the large improvement observed in the U.S. Virgin Islands may be influenced by limited data coverage or changes in monitoring practices over time.

In addition, aggregating data at the state and territory level can mask substantial variation within states, where counties may experience very different air quality trends. Finally, while this visualization identifies regions with improving or worsening air quality, it does not attribute these changes to specific pollutants, policies, or environmental factors.

Reflection

Next Steps

Future work should integrate air quality data with demographic and health information to better assess population level impacts and environmental equity. Analyzing annual data rather than only two time points would help capture short term variation and identify drivers of change. Expanding pollutant specific analysis could also improve understanding of regional air quality risks.

Source Files

Data Usage

A U.S. county GeoJSON file from Plotly was used to provide county boundaries for the choropleth map. This file was used only for visualization and does not contain air quality data.

State and county FIPS codes were obtained from the U.S. Census Bureau 2020 reference files. These codes were used to match air quality records to the correct counties so the data could be mapped accurately.